Professional Documents
Culture Documents
2022 03 21 A Simple Dynamic Model For Demand and Supply Shocks FINAL PDF
2022 03 21 A Simple Dynamic Model For Demand and Supply Shocks FINAL PDF
(GCD) model
for demand, supply and price shocks
Erhard Glötzl
Linz, Austria
erhard.gloetzl@gmail.com
February 2022
JEL: A12, B13, B41, B59, C02, C30, C54, C60, E10, E70
Abstract
In economics balance identities as e.g. C+K'-Y(L,K) = 0 must always apply.
Therefore, they are called constraints. This means that variables C,K,L cannot
change independently of each other. In General Equilibrium Theory (GE), the
solution for equilibrium is obtained as optimisation under the above or similar
constraints. The standard method for modelling dynamics in macroeconomics are
Dynamic Stochastic General Equilibrium (DSGE) models. Dynamics in DSGE
models result from the maximisation of an intertemporal utility function that
results in the Euler-Lagrange equations. The Euler-Lagrange equations are
differential equations that determine the dynamics of the system. In Glötzl, Glötzl,
und Richters (2019) we have introduced an alternative method to model
dynamics, which is constitutes a natural extension of GE theory. It is based on the
standard method for modelling dynamics under constraints in physics. We
therefore call models of this type "General Constrained Dynamic (GCD)" models.
GCD models can be seen as an alternative to DSGE models to model the dynamics
of economic processes. DSGE models are used in particular to analyse economic
shocks. For this reason, the aim of this article is to show how GCD models are
formulated and how they can be used to model economic shocks such as demand,
supply, and price shocks. Since the goal of this paper is to lay out the fundamental
principles to the formulation of such GCD models, very simple macroeconomic
models are used for illustrative purposes. All calculations can easily be carried
out with the open-source program GCDconfigurator, which also allows for the
integration of shocks.
Contents
Abstract.................................................................................................................. 2
Contents ................................................................................................................. 3
1. Introduction ..................................................................................................... 4
2. GCD models for non-intertemporal utility functions ..................................... 7
3. Modeling of supply, demand and price shocks............................................. 10
3.1. 2 different types of shocks ...................................................................... 10
3.2. Examples of demand shocks................................................................... 10
3.3. Examples of supply shocks ..................................................................... 11
3.4. Price shock .............................................................................................. 12
3.5. Policy shocks .......................................................................................... 13
4. Topics to be discussed................................................................................... 13
5. Model A1, (1 household, 1 firm, 1 good, without interest) .......................... 15
5.1. Overview of the setup ............................................................................. 15
5.2. Description of the A1 model in detail .................................................... 16
6. Calculations with model A1 on various shocks ............................................ 21
7. Calculations with model B1 for central bank polices in case of inflation and
deflation shock .................................................................................................... 29
7.1. Overview of the setup of model B1 ........................................................ 29
7.2. Inflation and deflation shock as variable shock for the price ................. 31
7.3. Model inflation and deflation shock as model shock ............................. 34
8. Summary ....................................................................................................... 35
Acknowledgements ............................................................................................. 35
References ........................................................................................................... 36
1. Introduction
https://github.com/lbinter/gcd
All Mathematica program codes used for calculations of the various GCD
models can be downloaded under
https://www.dropbox.com/sh/npis47xjqkecggv/AAAMzCVhmhDYIIhoB5MfA
TFya?dl=
DSGE models are typically used to analyse economic shocks. The target of this
article is to show how any kind of economic shock, e.g. demand, supply or price
shocks, can be modelled in the GCD framework. In this paper we limit ourselves
to GCD models with non-intertemporal utility functions.
In contrast to DSGE models, all previously published GCD models were based
on non-intertemporal utility functions. Since intertemporal utility functions are
essential in many applications and only intertemporal utility functions are used in
DSGE models, (Glötzl 2022a) describes the principles of formulating
intertemporal GCD models (IGCD) and shows that these IGCD models can be
seen as a generalisation of and alternative to DSGE models. Also in such GCD
models with intertemporal utility functions shocks can be integrated in the same
way as shown in the following for GCD models with non-intertemporal utility
functions.
In chapter 3 we describe the different types of economic shocks and how they can
be modelled.
In chapter 7 we describe the model B1, which includes a commercial bank and a
central bank to describe money creation. We present numerical calculations for
the impact of inflation and deflation shocks.
The GCD method is described in detail in Glötzl, Glötzl, und Richters (2019). We
will therefore limit ourselves to the explanation of a simple example with 2 agents
(each with 1 non-intertemporal utility function) and 1 constraint.
x U A
2
x2
The interest of A in a change of the variables does not lead alone to an actual
change, because the household must have also the power and/or possibility of
actually implementing its change desire. For example, a household cannot or can
only partially enforce its additional consumption desire, e.g., to go to the cinema
or go on vacation, because it is possibly quarantined or the borders are closed.
This limitation of the possibility to enforce his consumption change requests is
described by a (possibly time-dependent and endogenously determined) "power
factor" CH . In general, the change request for each of the variables is described by
"power factors" xH , xH . Considering the power factors, the following applies to
1 2
the change of x1 , x2 (due to the interest of A and the power of A to enforce this
interest)
A U A
x x1
x1
proportional to
1
x U A
2 xA2
x2
A U A B U B
x x1 x x1
1 x1
=
1
+ <2.1>
x A U A B U B
2 x
2 x x2
x2
2
If a constraint
Z ( x1 , x2 ) = 0
exists, there arises an additional constraint force f Z to the ex-ante force which
ensures that the constraint is fulfilled at all times. In physics, this constraint force
is perpendicular to the constraint at all times due to the so-called d'Alembert
principle, i.e.
Z ( x1 , x2 )
f ( x1 , x2 )
Z x1
f Z ( x1 , x2 ) = 1
= <2.2>
1
Z
f ( x1 , x2 ) Z ( x1 , x2 )
x2
The time-dependent factor = (t ) is called Lagrange multiplier, as in the case of
optimisation under constraints.
From <2.1> and <2.2> we get the equation of motion, which is called ex post
equation of motion:
A U A ( x1 , x2 ) B U B ( x1 , x2 ) Z ( x1 , x2 )
x
x1 x1 x1
x1 x1
J
U j ( x) K Z k ( x)
xi = xji + k i = 1,2,..., I <2.4>
j =1 xi k =1 xi
over time, i.e. → (t ) . For the sake of simplicity, we describe the temporal
behaviour of such a parameter (t ) by multiplication with a sawtooth curve:
(t ) = (t )
where the sawtooth curve is defined by
1 for t t s
f for t = t s
(t ) =
linear from f to 1 for t s t t s + d
1 for t s + d i j t
C → fCN C
At the same time, the constraint Z3 must always be fulfilled, even during
the shock. This is always guaranteed by the numerical solution method for
differential-algebraic equations of Mathematica NDSolve. In addition, of
course, one can make any other assumptions, such as that production Y and
investment K remain the same and that everything that is consumed less
(1 − f NC )C (t N ) is stored, i.e.
S (t A ) → S (t A ) + (1 − gCN )C (t A )
The dynamic system then continues to develop with these new initial
values.
→ (t ) = (t )
(3) Model shock: The power factor CH describes the power of the household
to actually enforce its consumption interests (e.g. due to quarantine
measures). A demand shock can be triggered by a change of CH in time
according to a sawtooth curve:
CH → CH (t ) = (t ) CH
For example, a supply shock A at the time can have the following causes:
(1) Model and variable shock: A supply shock could be triggered by the fact
that the production function
Y ( L, K ) = L K 1−
changes over time according to a sawtooth curve with a shock factor f A .
This initially corresponds to a model shock, because this is described by the
fact that the parameter changes according to a sawtooth curve with a
shock factor f A :
→ (t ) = (t )
At the same time, production Y suddenly changes by the shock factor f A at
the time of the shock t A :
Y (t A ) → f A Y (t A )
Because of the constraint
Z3 = 0 = Y ( L, K ) − C − K '− S '
at the time t A , therefore, C , K , S must also change so that the constraint is
fulfilled. This leads to sudden changes in at least one of the variables or in
all of them. This is always guaranteed by NDSolve. For example, one can
also make additional more precise assumptions about the behaviour of the
other variables, e.g. one could assume that at the time t A also C , K , S
change by the shock factor f A , i.e.
C (t A ) → f A C (t A ), K (t A ) → f A K (t A ), S (t A ) → f A S (t A )
and that the dynamic system can then adapt to these new starting conditions
and the time-varying parameter
(t ) = A (t )
(2) Model shock: The model parameter describes the labour intensity of
production. A supply shock can be triggered by the changes of over time
according to a sawtooth curve
→ (t ) = (t )
(3) Model shock: The power factor KF describes the power of the firm to
actually enforce its investment interests (e.g. because of administrative
regulations). A supply shock can be triggered by the fact that the power
factor KF changes over time according to a sawtooth curve:
KF → KF (t ) = (t ) KF
For example, a price shock P can be modelled by changing the price p at the time
t P by the factor f pP
p (t P ) → f pP p (t P )
This corresponds to a variable shock.
3.5. Policy shocks
In addition, a wide variety of fiscal and monetary policy measures that apply from
certain time points can of course also be interpreted as economic policy shocks
and modelled in the same way, e.g:
- government measures:
o Tax reform
o Increase or decrease in public debt
o etc.
- Changes in central bank policy:
o From money supply control to interest rate control
o Change in the inflation target
o Purchase programmes
o etc.
4. Topics to be discussed
From an economic point of view, there are 2 fundamental topics related to shocks:
(1) Forecasting: How will the economic variables change?
(2) Evaluating countermeasures: What measures can be taken to overcome the
shock as quickly as possible or with as little effort as possible?
Possible measures are, for example:
- Various forms of financial assistance from the government to firms
- Various forms of financial benefits to consumers
- Different ways of financing additional government expenditure
- short-time working models
- Central bank monetary policy measures
- Organisational measures, e.g. relieving companies of administrative
regulations, extending opening hours in the retail sector, etc. Such
organisational measures are expressed in the models by changes in power
factors or other parameters.
The target of section C. is to show that GCD models are basically suitable for
answering these 2 questions and that this can be done very easily and conveniently
with the help of the open-source program GCDconfigurator. The additions
necessary to incorporate shocks into a model programmed with GCDconfigurator
are very easy to program.
In order to apply GCD models to real economic situations, they would of course
have to be extended accordingly and adapted to the real conditions.
Using model A1 as an example, we show how special supply, demand and price
shocks can be modelled and what effects they have on the further course of the
economy.
Using model B1, we show how central bank measures have different effects on a
price shock depending on whether the central bank pursues a monetary policy or
an interest rate policy.
In order to make the effects clearly visible, the model calculations are carried out
for very strong shocks of a magnitude that is unlikely to occur in reality.
5. Model A1, (1 household, 1 firm, 1 good, without
interest)
In addition, one can be interested, for example, in net investment, for which one
defines as a further algebraically defined variable
inv ( K ) = K ' <5.2>
Households want to consume with decreasing marginal utility. Consumption of
consumer goods C leads to a utility for households in the amount of C with
0 1 . They strive for a desired working time L̂ . Deviations from the desired
working time L̂ lead to a reduction of utility by ( L − Lˆ ) 2 . In addition, households
aim to keep cash in the amount of Mˆ H . Deviations from the desired cash position
Mˆ H lead to a reduction in utility by ( Mˆ H − M H ) 2 . This leads to the utility function
for the household
U H = C − ( Lˆ − L) 2 − ( Mˆ H − M H ) 2 0 1 <5.3>
For the company, in the simplest case, the utility initially consists of the goods
produced, which are valued at the selling price, i.e. pY . The produced goods are
used for:
C Sales = Consumption
S change in inventory
K changes in productive capital stock
In principle, it would be possible to weight the utility of these uses differently.
For the sake of simplicity, we will refrain from doing so. Therefore, this utility is
reduced by the cost of labor and the cost of storage, which we evaluate through
the deviations from the planned inventory. For simplicity, we assume that holding
money in cash has no influence on the utility. This leads to the utility function
for the firm
U F = pY ( L, K ) − w L − ( Sˆ − S ) 2 = p L K 1− − w L − (Sˆ − S ) 2 <5.4>
From the model graph, it can be seen that the following constraints must be
satisfied:
Z1 = 0 = w L − p C − M H ' for money of household H
Z2 = 0 = p C − w L − M F ' for money of firm F <5.5>
Z 3 = 0 = Y ( L, K ) − C − K '− S ' for good 1 of firm F
According to the methodology of GCD models, the interest or desire of
households to change consumption is the greater the more the utility changes
U H
when consumption changes, i.e., the interest is proportional to . However,
C
the interest in changing consumption does not in itself lead to an actual change in
consumption, because the household must also have the power or opportunity to
actually implement its desire to change consumption. For example, a household
cannot or can only partially enforce its additional consumption wish, e.g., to go to
the cinema or on holiday, because it is in quarantine or the borders are closed.
This restriction of the possibility to enforce his or her consumption change wishes
is described by a (possibly time-dependent) "power factor" CH . Analogously, the
U F
firm could have an interest and power CF to influence consumption. In the
C
U F
specific case = 0 . This results in the following behavioural equation for the
C
ex-ante planned change in consumption
U H U F
C ' = CH + CF = CH C −1
C C
<5.6>
The same considerations apply to labour L as to consumption. Even the
household's wish to increase or reduce working time does not in itself lead to an
actual change in working time, because the household must also have the power
or possibility to actually implement its wish to change. For example, a household
might not be able to enforce its wish to increase working time, or only partially,
because it is on short-time working or unemployed, or it might not be able to
enforce its wish to reduce working time because it is contractually obliged to work
overtime. This restriction of the possibility to enforce his wishes for a change in
working time is also described by a (possibly time-dependent) power factor,
which we denote with LH . The same applies to the firm's ability to influence
working time.
Therefore, the behavioural equation for the ex-ante planned change in working
time is as follows
U H F U
F
L' = H
+ L = 2 LH ( Lˆ − L) + LF ( p L −1K 1− − w )
L L
L
The ex-ante behavioural equations for the other variables result analogously.
However, the plans of the 2 agents household and firm to change consumption C,
labour L and the other variables cannot be enforced independently of each other,
because the constraints
Z1 = 0 = w L − p C − M H ' für Geld von Haushalt H
Z2 = 0 = p C − w L − M ' F
für Geld von Firma F <5.7>
Z 3 = 0 = Y ( L, K ) − C − K '− S ' − DP für Gut 1 von Firma F
lead to constraint forces, which we assume are vertical constraint forces. The
constraint force for the change in consumption therefore results in
Z1 Z Z
1 + 2 2 + 3 3 = − 1 p + 2 p − 3
C C C
The behavioural equation for the actual ex-post change in consumption is
therefore
U H Z Z Z
C ' = CH + 1 1 + 2 2 + 3 3 = CH C −1 − 1 p + 2 p − 3 <5.8>
C C C C
Analogously, the actual ex-post change in labour is as follows
U H U F Z Z Z
L ' = LH + LF + 1 1 + 2 2 + 3 3 =
L L L L L
H ˆ
= 2 L ( L − L) + L ( p L K − w ) + 1w − 2 w + 3 L −1K 1−
F −1 1−
This also applies analogously to the company's investments. In the case of the
company, too, the actual implementation of ex-ante planned investment increases
can be prevented by real restrictions, e.g. by interruptions in supply chains. In the
same way, a desired reduction in investment may not be possible to the desired
extent because the project is a large-scale project of many years' duration. These
restrictions can in turn be described by a (possibly time-dependent) power factor
KB . This results in the following behavioural equation for the actual ex-post
change in capital
U F Z Z Z
K ' = KF + 1 1 + 2 2 + 3 3 = KF p (1 − ) L K − − 3 <5.9>
K K K K '
Z Z
Note that we have to use 3 instead of 3 because the constraint forces are
K ' K
always derived from the highest time derivative of the variables (see Remark in
chapter 2 (Flannery 2011).
The equations of behaviour for M H , M F , S , p, w are derived analogously. In sum,
this results in the model equations
differentiell behavioural equations
U H U F Z Z Z
C ' = CH + CF + 1 1 + 2 2 + 3 3 =
C C C C C
−1
= C C − 1 p + 2 p − 3
H
U H U F Z Z Z
L ' = LH + LF + 1 1 + 2 2 + 3 3 =
L L L L L
H ˆ −1 1−
= L ( L − L) + 1w − 2 w + 3 L K
U H U F Z Z Z
K ' = KH + KF + 1 1 + 2 2 + 3 3 =
K K K K K '
−
= K p (1 − ) L K − 3
F
U H U F Z1 Z 2 Z 3
M H ' = MH H + F
+ 1 + 2 + 3 =
M H MH
M H
M '
H
M H
M H
(
= 2 H H Mˆ H − M H − H
M )
U U F
H
H Z
H
B Z
B
Z1
M '=
F H
+ M F
F
+ + + 1 =
MF
M F
M F
M '
F
M F
M F
= −2
U H U F Z Z Z
S ' = SH + SF + 1 1 + 2 2 + 3 3 =
S S S S S '
(
= F Sˆ − S −
S ) 3
U H
U Z Z
F
Z
p ' = pH + pF + 1 1 + 2 2 + 3 3 =
p p p p p
= pF K 1− L − 1c + 2c
U H U F Z Z Z
w ' = wH + wF + 1 1 + 2 2 + 3 3 =
w w w w w
= − w L + 1L − 2 L
F
(
M H ' = 2 MH H Mˆ H − M H − 1 )
M F ' = −2
(
S ' = SF Sˆ − S − 3 )
p ' = pF K 1− L − 1c + 2c
w ' = − wF L + 1L − 2 L
6. Calculations with model A1 on various shocks
With no shocks
With multiple shocks
Price shock at t = 20, p→2p
Demand shock due to variable shock (consumption shock)
at t = 15 C → 0.5 C
Demand shock due to model shock (shock to power of households)
at t = 10 CH → 0.5CH thereafter, the power of the households increases again
linearly to CH within the time period dn = 5
Supply shock due to model shock (shock to power of the firm)
at t = 10 KF → 0.5 KF thereafter, the power of the firm increases again linearly
to KF within the time period da = 5
The target of model B1 is to model the money creation process by the central bank
in a simplified way.
In model B1, the central bank is seen as an endogenous money creator and the
bank is seen as an endogenous credit creator. The central bank's target is to keep
p
inflation at the target inflation pˆ = 0.02 i.e. 2% by means of interest rate policy
p
( = 1 ) and monetary-supply policy( = 0 ).
In this model B1, the central bank's interest rate policy is still modeled in a very
simplified way. We assume that the policy rate is constant 0 (banks do not pay
interest to the central bank) and that the central bank can, however, influence the
interest rate directly. That the policy rate is constant 0 is possible and does not
cause the bank to borrow arbitrarily from the central bank, since the bank is
assumed to have a constant 0 utility function. This means that the bank has no
particular interest in lending to firms or receiving savings deposits from
households. Thus, the bank lends endogenously and accepts savings deposits
endogenously.
https://www.dropbox.com/s/th868lbth9ahz9i/Modell%20B1%20SCHOCK%20
Version%203.ndsolve.nb?dl=0
Certainly, one could also interpret these calculations in economic terms. But
without prior adjustment of the models to real conditions, a real interpretation is
not really serious. Therefore, we will not comment further on the calculated
graphs.
As emphasized several times, the target of this book is to present the methodology
of the GCD models in principle and to give an idea of what can be done with them
and in what form. For application to concrete economic questions, the GCD
models still need to be adapted to real conditions. This is one of the tasks that still
has to be done in the future.
Model B1 without shock
Inflation shock p → 1.5 p , pure money supply policy of central bank = 0
8. Summary
GCD models are a natural extension of GE theory. They are based on the standard
method for modelling dynamics under constraints in physics and can be thought
of as an alternative to DSGE models. DSGE models are typically used to analyse
economic shocks. For this reason, the aim of this article was to show how any
kind of economic shock, e.g. demand, supply or price shocks, can be modelled
within the framework of GCD models.
Acknowledgements
My thanks go above all to my son Florentin, who has advised me on all economic
matters over the years, and especially to Oliver Richters, who has been a faithful
companion in all matters of economics and physics for many years and without
whom many of the publications would not have been possible.
References
Flannery, Martin Raymond. 2011. „D’Alembert–Lagrange analytical dynamics
for nonholonomic systems“. Journal of Mathematical Physics 52 (3):
032705. https://doi.org/10.1063/1.3559128.
Glötzl, Erhard. 2015. „Why and How to overcome General Equilibrium
Theory“. MPRA Paper 66265. https://mpra.ub.uni-
muenchen.de/66265/1/MPRA_paper_66265.pdf.
———. 2022a. „General Constrained Dynamic (GCD) models with
intertemporal utility functions“.
https://www.dropbox.com/s/269ki6287fc8atk/A%20general%20constrain
ed%20dynamic%20model%20with%20intertemporal%20utility%20functi
on%20FINAL%20Version%202%20maten.docx?dl=0.
———. 2022b. „Macroeconomic General Constrained Dynamic models (GCD
models)“.
https://www.dropbox.com/s/2h8o3rg0r4b008b/Macroeconomic%20GCD
%20Modelle%20A%2CB%2CC%20FINAL%20Version%202.docx?dl=0
.
Glötzl, Erhard, und Lucas Binter. 2022. „GCDconfigurator, GitHub“.
https://github.com/lbinter/gcd.
Glötzl, Erhard, Florentin Glötzl, und Oliver Richters. 2019. „From Constrained
Optimization to Constrained Dynamics: Extending Analogies between
Economics and Mechanics“. Journal of Economic Interaction and
Coordination 14 (3): 623–42. https://doi.org/10.1007/s11403-019-00252-
7.
Godley, Wynne, und Marc Lavoie. 2012. Monetary economics: an integrated
approach to credit, money, income, production and wealth. 2. Aufl.
Basingstoke and New York: Palgrave Macmillan.
Richters, Oliver. 2021. „Modeling the Out-of-Equilibrium Dynamics of
Bounded Rationality and Economic Constraints“. Journal of Economic
Behavior & Organization 188 (August): 846–66.
https://doi.org/10.1016/j.jebo.2021.06.005.
Richters, Oliver, und Erhard Glötzl. 2020. „Modeling Economic Forces, Power
Relations, and Stock-Flow Consistency: A General Constrained
Dynamics Approach“. Journal of Post Keynesian Economics 43 (2): 281–
97. https://doi.org/10.1080/01603477.2020.1713008.
Smale Institute. 2003. „Smale Institute Mathematics & Computation“. 2003.
http://www.smaleinstitute.com/problem.html.
Smale, Steve. 1991. „Dynamics Retrospective: Great Problems, Attempts That
Failed“. Physica D: Nonlinear Phenomena 51 (1–3): 267–73.
https://doi.org/10.1016/0167-2789(91)90238-5.
———. 1997. „Mathematical Problems for the Next Century“. Fields Institute,
Toronto, Juni.
http://www.cityu.edu.hk/ma/doc/people/smales/pap104.pdf.
———. 1998. „Mathematical problems for the next century“, nachgedruckt in
V. Arnold, M. Atiyah, P. Lax, B. Mazur: Mathematics: Frontiers and
Perspectives 2000, American Mathematical Society 2000, 2: 7–15.
Zaman, Asad. 2020. „New Directions in Macroeconomics“. International
Econometric Review (IER) 12 (1): 1–23.
https://doi.org/10.33818/ier.747603.